INTRODUCTION

‘We would maintain that, for any cognitive capacity, demonstrating that that capacity is cognitively penetrable indicates that the capacity derives from an information base rather than from an off-line simulation’. So write Nichols, Stich, Leslie and Klein (1996), echoing some earlier claims of Stich and Nichols (1995a, 1995b). But are they right? I shall argue first that they are not, and that a simulationist may admit the possibility of cognitive penetration. This discussion will comprise the next three sections of the essay, Section 2 setting the scene and Sections 3 and 4 presenting the main considerations against the quoted claim.

But following that I shall ask whether cognitive penetration of the kind outlined is, in fact, likely to be the source of error in the cases cited by opponents of simulation, and I shall suggest that it is not. In understanding the causes of these errors it is probably more important to grasp what simulation can and cannot plausibly claim to deal with and certain potential complexities of its operation. The domain of simulation is considered in Section 5, where I urge that rational connection of content is the matter that simulation should centrally be expected to cope with. Section 6 summarises possible sources of error that a simulationist can acknowledge without embarrassment and offers some comments on the errors discussed in the literature.